14:05:32scymtymdo we routinely produce stack configurations that SB-DI:X86-CALL-CONTEXT cannot decode? (sb-sprof:with-profiling () (asdf:load-system :ironclad :force t)) seems to always result in a couple of interruptions for which SB-DI:X86-CALL-CONTEXT fails on the first frame, resulting in a bogus trace (unless i heuristically ignore "implausible" traces)
14:06:42scymtymmuch more rarely, the sigprof handler thinks the control stack pointer is invalid right from the start (why does the current implementation record a stray sample in that case?)